To configure your artificial intelligence agents, access the "Prompt and AI" module in the side menu of the Papo platform (https://papo.global/). In this area, you can create and edit the prompts that guide AI responses, define behavior rules, assign agents to specific channels, and analyze usage statistics to optimize your customer service automation.
The Prompt and AI module centralizes control over how your artificial intelligence interacts with customers. You can have a main agent, which serves as the default for the entire account, and create sub-assistants with specific instructions for different products, channels, or service phases.
The table below summarizes the main configuration areas available in the module:
| Configuration Area | What you can do |
|---|---|
| Agent List | View the "General" agent (default) and all created sub-assistants. |
| Prompt Editor | Define the instructions the AI will follow. This can be free text or divided into sections. |
| General Rules | Establish behavior guidelines that apply to all agent interactions. |
| AI Settings | Make technical adjustments to the AI model, such as response creativity. |
| Inboxes | Enable or disable AI for a specific channel (e.g., WhatsApp, Instagram) and choose which agent will respond. |
| Statistics | Monitor AI performance, with data on messages sent, failures, and skipped interactions. |
How to Access the Prompt and AI Module
To start configuring your assistants, follow these steps:
- Log in to your Papo account.
- In the left-hand side menu, locate and click on Prompt and AI.
If the item is not visible in your menu, there might be two reasons:
- Module disabled: The module might not be active for your account.
- Lack of permission: Your account administrator might have restricted your access using custom roles. In this case, ask the administrator to review your permissions.
Step-by-Step Guide to Configure an AI Agent
Within the module, you will find the "General" agent and the option to create new sub-assistants. The configuration process is the same for both.
1. Select or Create an Agent
You can edit the "General" agent, which functions as the default, or click to add a new sub-assistant. New sub-assistants are created as inactive drafts, allowing you to configure them completely before activating them.
2. Define the Instructions (Prompt)
This is the most important step. The prompt is the "brain" of your agent. You can provide instructions in two ways:
- Free Text: Write a detailed paragraph describing who the AI is, its objective, the tone of voice it should use, what it can and cannot do.
- Structured Instructions: Fill in up to seven predefined sections to better organize the guidelines, covering aspects such as persona, objectives, restrictions, and response examples.
A good prompt is clear, specific, and provides enough context for the AI to respond accurately and in line with your brand. You can also preview how the final prompt will be interpreted by the system.
3. Adjust General Rules and Settings
In the agent's settings area, you can define general rules that apply to all its interactions and adjust AI model parameters. The platform comes with default settings that work well for most cases, but advanced users can make fine adjustments.
4. Activate the Agent in an Inbox
A configured agent, even if saved, will only take action when associated with one or more inboxes.
- Go to the Inboxes tab or section within the Prompt module.
- You will see a list of your connected channels (WhatsApp, Instagram, etc.).
- For each channel, you can turn the AI on or off.
- When turning it on, select which agent (the General or a sub-assistant) should be responsible for that channel.
This allows, for example, having a sales-specialized agent responding on WhatsApp and another focused on support responding to Instagram comments.
Analyzing Usage Statistics
To understand if your AI is working as expected, access the Statistics area. There, you will find graphs and numbers about AI interactions, including:
- Succeeded: The number of messages the AI successfully sent.
- Failed: Interactions where the AI attempted to respond but an error occurred.
- Skipped: Situations where the AI decided not to respond, either due to a rule or a lack of confidence in the answer.
Analyzing this data helps identify the need to adjust prompts, rules, or the knowledge base that feeds the AI.
Frequently Asked Questions
What is the difference between the main agent and sub-assistants?
The "General" or main agent is the default for your account. If you activate AI in an inbox without specifying an agent, it will be the one to respond. Sub-assistants are specialized agents you can create for specific tasks (e.g., lead qualification, technical support) and assign to distinct channels or situations.
What happens if I don't configure an AI agent?
If you don't edit the prompt, the agent will use a set of default Papo instructions, which are generic. To get personalized responses aligned with your business, it is crucial that you edit the "General" agent's prompt with your own information.
Can I have a different AI for each channel (WhatsApp, Instagram)?
Yes. In the "Prompt and AI" module, you can access the configuration of your inboxes and assign a specific agent to each one. You can use the "General" agent for one channel and a specialized sub-assistant for another.
How do I know if my AI is responding correctly?
The best way is to monitor conversations in the inbox and analyze the "Statistics" section in the Prompt module. Check the success rates and review conversations where the AI interacted to ensure that the responses are aligned with your instructions.
I can't see the "Prompt and AI" module. What should I do?
First, check with your Papo account administrator if the module is enabled for the company and if your user profile has permission to access it. Access to advanced modules is managed through custom roles.
Read Also
- Feeding the AI: How to Create a Knowledge Base for Automatic Responses
- Creating your first Automation Flow from scratch
The first user is free, no credit card required.